From 494af585b04d55d8153b75b9f19c6d3f7dffe8d3 Mon Sep 17 00:00:00 2001
From: Stefan Brass <stefan.brass@informatik.uni-halle.de>
Date: Fri, 11 Oct 2019 10:53:36 +0200
Subject: [PATCH] Improved runtime estimation tables/views
---
db/TODO | 2 ++
db/drop_db.sql | 3 ++-
db/estimate_par.sql | 41 +++++++-----------------------
db/estimate_par_old.sql | 55 +++++++++++++++++++++++++++++++++++++++++
db/load_data | 5 +++-
db/v_check.sql | 18 ++++++++++++--
db/v_estimate.sql | 26 +++++--------------
graph/bench_cost.sql | 45 ---------------------------------
8 files changed, 94 insertions(+), 101 deletions(-)
create mode 100644 db/estimate_par_old.sql
delete mode 100644 graph/bench_cost.sql
diff --git a/db/TODO b/db/TODO
index a40e07d..f8c6588 100644
--- a/db/TODO
+++ b/db/TODO
@@ -10,3 +10,5 @@ Columns TC_COST etc. removed from table INPUT_GRAPHS
Probably, it would be possible new to merge TCFF_PLOT_READ and SGFF_PLOT_REAL
and so on (the new table BENCH_COST is more general than previous distinct
columns).
+
+New view INTEGRITY_ERROR
diff --git a/db/drop_db.sql b/db/drop_db.sql
index 66224b7..82d0d29 100644
--- a/db/drop_db.sql
+++ b/db/drop_db.sql
@@ -2,7 +2,7 @@
-- Project: ddbench - Logic Programming and Database Benchmarks
-- Filename: db/drop_db.sql
-- Purpose: Database for Benchmark Runs: Drop all Tables and Views
--- Last Change: 04.10.2019
+-- Last Change: 11.10.2019
-- Language: SQL (Tested with PostgreSQL)
-- Author: Stefan Brass
-- EMail: brass@informatik.uni-halle.de
@@ -160,6 +160,7 @@ DROP VIEW DATASCRIPT;
-- Views for Detecting Outliers and other Data Checks (File v_check.sql):
-- ============================================================================
+DROP VIEW INTEGRITY_ERROR;
DROP VIEW PCT_CPU;
DROP VIEW PCT_CPU_INPUT;
DROP VIEW REAL_TOO_LARGE;
diff --git a/db/estimate_par.sql b/db/estimate_par.sql
index 644a2f3..fbb5a1f 100644
--- a/db/estimate_par.sql
+++ b/db/estimate_par.sql
@@ -2,7 +2,7 @@
-- Project: ddbench - Deductive System and Database Benchmarks
-- Filename: db/estimate_par.sql
-- Purpose: Database for Benchmark Runs: Data for Runtime Estimation
--- Last Change: 02.02.2019
+-- Last Change: 10.10.2019
-- Language: SQL (Tested with PostgreSQL)
-- Author: Stefan Brass
-- EMail: brass@informatik.uni-halle.de
@@ -21,35 +21,12 @@
-- Columns: SYS, BENCH, STARTUP, LOAD_IDX, RULE_APP, RESULT, HITS
-- START LOAD R_APP RESULT HITS
-INSERT INTO ESTIMATE_PAR VALUES( 'xsb', 'tcff', 125, 100, 107, 390, 41);
-
-INSERT INTO ESTIMATE_PAR VALUES( 'yap', 'tcff', 30, 72, 263, 350, 32);
-
-INSERT INTO ESTIMATE_PAR VALUES( 'bam', 'tcff', 0, 244, 35, 51, 29);
-
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1510, 2030, 164, 1590, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1605, 1140, 174, 1578, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1705, 1130, 172, 1578, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1805, 1130, 172, 1566, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1905, 1120, 172, 1566, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2005, 1120, 172, 1554, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2105, 1110, 172, 1554, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2205, 1110, 170, 1548, 42);
-INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2305, 1100, 172, 1542, 42);
-
-INSERT INTO ESTIMATE_PAR VALUES('psql', 'tcff', 75, 570, 440, 1020, 38);
-INSERT INTO ESTIMATE_PAR VALUES('psql', 'tcff', 105, 180, 446, 1116, 38);
-
-INSERT INTO ESTIMATE_PAR VALUES('sqlite3','tcff', 50, 380, 643, 1524, 42);
-INSERT INTO ESTIMATE_PAR VALUES('sqlite3','tcff', 105, 0, 628, 1818, 42);
-
-INSERT INTO ESTIMATE_PAR VALUES('mariadb','tcff', 0, 1210, 422, 1362, 19);
--- InnoDB: 0 1210 422 1362 19
--- Memory: 5, 410, 507, 822, 18
-
-INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 96, 68, 6, 909, 24);
-INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 100, 40, 7, 906, 24);
-INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 100, 26, 7, 909, 24);
-INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 105, 30, 7, 894, 24);
-INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 120, 26, 7, 897, 24);
+INSERT INTO ESTIMATE_PAR VALUES( 'xsb', 'tcff', 130, 90, 95, 284, 49);
+INSERT INTO ESTIMATE_PAR VALUES( 'yap', 'tcff', 20, 4, 200, 221, 35);
+INSERT INTO ESTIMATE_PAR VALUES( 'bam', 'tcff', 0, 373, 32, 54, 31);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2076, 1599, 166, 1402, 47);
+INSERT INTO ESTIMATE_PAR VALUES('psql', 'tcff', 369, 102, 419, 711, 48);
+INSERT INTO ESTIMATE_PAR VALUES('sqlite3','tcff', 90, 93, 593, 1317, 49);
+INSERT INTO ESTIMATE_PAR VALUES('mariadb','tcff', 19, 1026, 438, 1164, 24);
+INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 25, 12, 42, 671, 32);
diff --git a/db/estimate_par_old.sql b/db/estimate_par_old.sql
new file mode 100644
index 0000000..644a2f3
--- /dev/null
+++ b/db/estimate_par_old.sql
@@ -0,0 +1,55 @@
+-- ============================================================================
+-- Project: ddbench - Deductive System and Database Benchmarks
+-- Filename: db/estimate_par.sql
+-- Purpose: Database for Benchmark Runs: Data for Runtime Estimation
+-- Last Change: 02.02.2019
+-- Language: SQL (Tested with PostgreSQL)
+-- Author: Stefan Brass
+-- EMail: brass@informatik.uni-halle.de
+-- WWW: http://www.informatik.uni-halle.de/~brass/
+-- Address: Feldschloesschen 15, D-06120 Halle (Saale), GERMANY
+-- Copyright: (c) 2016-2019 by Stefan Brass
+-- License: See file "LICENSE" for copying conditions.
+-- Note: There is no warranty at all - this code may contain bugs.
+-- ============================================================================
+
+
+-- ============================================================================
+-- Parameter Values for Runtime Estimation (Experimental Table):
+-- ============================================================================
+
+-- Columns: SYS, BENCH, STARTUP, LOAD_IDX, RULE_APP, RESULT, HITS
+
+-- START LOAD R_APP RESULT HITS
+INSERT INTO ESTIMATE_PAR VALUES( 'xsb', 'tcff', 125, 100, 107, 390, 41);
+
+INSERT INTO ESTIMATE_PAR VALUES( 'yap', 'tcff', 30, 72, 263, 350, 32);
+
+INSERT INTO ESTIMATE_PAR VALUES( 'bam', 'tcff', 0, 244, 35, 51, 29);
+
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1510, 2030, 164, 1590, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1605, 1140, 174, 1578, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1705, 1130, 172, 1578, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1805, 1130, 172, 1566, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 1905, 1120, 172, 1566, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2005, 1120, 172, 1554, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2105, 1110, 172, 1554, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2205, 1110, 170, 1548, 42);
+INSERT INTO ESTIMATE_PAR VALUES('jena', 'tcff', 2305, 1100, 172, 1542, 42);
+
+INSERT INTO ESTIMATE_PAR VALUES('psql', 'tcff', 75, 570, 440, 1020, 38);
+INSERT INTO ESTIMATE_PAR VALUES('psql', 'tcff', 105, 180, 446, 1116, 38);
+
+INSERT INTO ESTIMATE_PAR VALUES('sqlite3','tcff', 50, 380, 643, 1524, 42);
+INSERT INTO ESTIMATE_PAR VALUES('sqlite3','tcff', 105, 0, 628, 1818, 42);
+
+INSERT INTO ESTIMATE_PAR VALUES('mariadb','tcff', 0, 1210, 422, 1362, 19);
+-- InnoDB: 0 1210 422 1362 19
+-- Memory: 5, 410, 507, 822, 18
+
+INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 96, 68, 6, 909, 24);
+INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 100, 40, 7, 906, 24);
+INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 100, 26, 7, 909, 24);
+INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 105, 30, 7, 894, 24);
+INSERT INTO ESTIMATE_PAR VALUES('souffle','tcff', 120, 26, 7, 897, 24);
+
diff --git a/db/load_data b/db/load_data
index 03c1a47..3a34893 100755
--- a/db/load_data
+++ b/db/load_data
@@ -3,7 +3,7 @@
# Project: ddbench - Deductive System and Database Benchmarks
# Filename: db/load_data
# Purpose: Shellscript to create database and load all data
-# Last Change: 04.10.2019
+# Last Change: 10.10.2019
# Language: Shellscript (bash)
# Author: Stefan Brass
# EMail: brass@informatik.uni-halle.de
@@ -98,6 +98,9 @@ psql -f input_join1.sql | grep -v "INSERT 0 1"
# Cost measures of a benchmark for an input file:
psql -f bench_cost.sql | grep -v "INSERT 0 1"
+# For runtime estimation:
+psql -f estimate_par.sql | grep -v "INSERT 0 1"
+
# Convert tsv-files with benchmark results to SQL:
# (might be unnecessary, if the SQL files already exist, but is safer,
# because SQL files might be missing or outdated)
diff --git a/db/v_check.sql b/db/v_check.sql
index b0f1def..ba70184 100644
--- a/db/v_check.sql
+++ b/db/v_check.sql
@@ -2,7 +2,7 @@
-- Project: rbench - Logic Programming and Database Benchmarks
-- Filename: db/v_check.sql
-- Purpose: Benchmark DB: Views for Outlier Detection and other Checks
--- Last Change: 04.03.2019
+-- Last Change: 10.10.2019
-- Language: SQL (Tested with PostgreSQL)
-- Author: Stefan Brass
-- EMail: brass@informatik.uni-halle.de
@@ -15,7 +15,7 @@
-- ############################################################################
--- Outlier Detection, Checking of Results for Plausibility:
+-- Outlier Detection, Checking of Results for Plausibility, Integrity Check:
-- ############################################################################
@@ -141,3 +141,17 @@ AND WITH_OUTPUT = 'N'
GROUP BY SYS, MACHINE, BENCH, FILE_ID
ORDER BY SYS, MACHINE, BENCH, FILE_ID;
+
+-- ============================================================================
+-- Integrity Constraints / Consistency Check - Query Result must be Empty!
+-- ============================================================================
+
+CREATE VIEW INTEGRITY_ERROR AS
+SELECT 'Number of lines in INPUT_FILE and edges in INPUT_GRAPH differ for: ' ||
+ F.FILE_ID
+FROM INPUT_FILE F, INPUT_GRAPH G
+WHERE F.FILE_ID = G.FILE_ID
+AND F.LINES <> G.NUM_EDGEs;
+
+-- Further checks can be added with UNION ALL
+
diff --git a/db/v_estimate.sql b/db/v_estimate.sql
index 9c1a50d..ff87e66 100644
--- a/db/v_estimate.sql
+++ b/db/v_estimate.sql
@@ -2,7 +2,7 @@
-- Project: rbench - Logic Programming and Database Benchmarks
-- Filename: db/v_estimate.sql
-- Purpose: Benchmark DB: Views to Support Research on Runtime Estimation
--- Last Change: 04.10.2019
+-- Last Change: 11.10.2019
-- Language: SQL (Tested with PostgreSQL)
-- Author: Stefan Brass
-- EMail: brass@informatik.uni-halle.de
@@ -73,13 +73,13 @@ CREATE VIEW ESTIMATE_FORMULA AS
SELECT R.SYS, E.STARTUP, E.LOAD_IDX, E.RULE_APP, E.RESULT,
G.FILE_ID,
ROUND(CAST(
- E.STARTUP +
+ E.STARTUP/1000 +
(G.NUM_EDGES * (LOG(G.NUM_EDGES)/LOG(2)) * E.LOAD_IDX / 1000000000) +
(C.INST * E.RULE_APP / 1000000000) +
(C.SIZE * E.RESULT / 1000000000) AS NUMERIC),
3) AS ESTIMATE,
R.REAL_T
-FROM RESULT R, INPUT_GRAPH G, BENCH_COST C, ESTIMATE_PAR E
+FROM BEST_RESULT R, INPUT_GRAPH G, BENCH_COST C, ESTIMATE_PAR E
WHERE R.SYS = E.SYS AND E.BENCH = 'tcff' AND R.BENCH = 'tcff'
AND C.BENCH = 'tcff' AND C.FILE_ID = G.FILE_ID
AND R.FILE_ID = G.FILE_ID
@@ -144,24 +144,10 @@ SELECT R.SYS, R.BENCH, R.IMPL, F.FILE_ID, F.SEQ_NO,
C.ITER || '","' ||
ROUND(R.REAL_T*1000) || '"' AS CSV
FROM INPUT_GRAPH G, INPUT_FILE F, BENCH_COST C, BEST_RESULT R
-WHERE R.BENCH = 'tcff' AND G.FILE_ID = F.FILE_ID AND R.FILE_ID = F.FILE_ID
-AND C.BENCH = 'tcff' AND C.FILE_ID = F.FILE_ID
-UNION ALL
-SELECT R.SYS, R.BENCH, R.IMPL, F.FILE_ID, F.SEQ_NO,
- '"' || R.SYS || '","' ||
- R.BENCH || '","' ||
- R.IMPL || '","' ||
- G.FILE_ID || '","' ||
- G.NUM_NODES || '","' ||
- G.NUM_EDGES || '","' ||
- C.SIZE || '","' ||
- C.INST || '","' ||
- C.ITER || '","' ||
- ROUND(R.REAL_T*1000) || '"' AS CSV
-FROM INPUT_GRAPH G, INPUT_FILE F, BENCH_COST C, BEST_RESULT R
-WHERE R.BENCH = 'sgff' AND R.FILE_ID = F.FILE_ID
+WHERE R.BENCH = C.BENCH
+AND R.FILE_ID = F.FILE_ID
AND G.FILE_ID = F.FILE_ID
-AND C.BENCH = 'sgff' AND C.FILE_ID = F.FILE_ID
+AND C.FILE_ID = F.FILE_ID
UNION ALL
SELECT '' AS SYS, '' AS BENCH, '' AS IMPL, '' AS FILE_ID, 0 AS SEQ_NO,
'"' || 'SYS' || '","' ||
diff --git a/graph/bench_cost.sql b/graph/bench_cost.sql
deleted file mode 100644
index 686d8e2..0000000
--- a/graph/bench_cost.sql
+++ /dev/null
@@ -1,45 +0,0 @@
-INSERT INTO BENCH_COST VALUES('tcff','b17',1966082,16,1966082);
-INSERT INTO BENCH_COST VALUES('tcff','b18',4194306,17,4194306);
-INSERT INTO BENCH_COST VALUES('tcff','c1k',1000000,1000,1001000);
-INSERT INTO BENCH_COST VALUES('tcff','c2k',4000000,2000,4002000);
-INSERT INTO BENCH_COST VALUES('tcff','c3k',9000000,3000,9003000);
-INSERT INTO BENCH_COST VALUES('tcff','c4k',16000000,4000,16004000);
-INSERT INTO BENCH_COST VALUES('tcff','k100',10000,1,1010000);
-INSERT INTO BENCH_COST VALUES('tcff','k1k',1000000,1,1001000000);
-INSERT INTO BENCH_COST VALUES('tcff','k1k5',2250000,1,3377250000);
-INSERT INTO BENCH_COST VALUES('tcff','k2k',4000000,1,8004000000);
-INSERT INTO BENCH_COST VALUES('tcff','k50',2500,1,127500);
-INSERT INTO BENCH_COST VALUES('tcff','k500',250000,1,125250000);
-INSERT INTO BENCH_COST VALUES('tcff','m4_2ki',12288,3,12288);
-INSERT INTO BENCH_COST VALUES('tcff','m16_512',61440,15,61440);
-INSERT INTO BENCH_COST VALUES('tcff','m64_128',258048,63,258048);
-INSERT INTO BENCH_COST VALUES('tcff','m256_32',1044480,255,1044480);
-INSERT INTO BENCH_COST VALUES('tcff','m1ki_8',4190208,1023,4190208);
-INSERT INTO BENCH_COST VALUES('tcff','m4ki_2',16773120,4095,16773120);
-INSERT INTO BENCH_COST VALUES('tcff','p1k',499500,999,499500);
-INSERT INTO BENCH_COST VALUES('tcff','p2k',1999000,1999,1999000);
-INSERT INTO BENCH_COST VALUES('tcff','p3k',4498500,2999,4498500);
-INSERT INTO BENCH_COST VALUES('tcff','p4k',7998000,3999,7998000);
-INSERT INTO BENCH_COST VALUES('tcff','s1k_1',1000000,500,2002000);
-INSERT INTO BENCH_COST VALUES('tcff','s1k_3',1000000,250,4004000);
-INSERT INTO BENCH_COST VALUES('tcff','s1k_4',1000000,200,5005000);
-INSERT INTO BENCH_COST VALUES('tcff','s2k_1',4000000,1000,8004000);
-INSERT INTO BENCH_COST VALUES('tcff','s2k_3',4000000,500,16008000);
-INSERT INTO BENCH_COST VALUES('tcff','s2k_4',4000000,400,20010000);
-INSERT INTO BENCH_COST VALUES('tcff','t50',1225,1,20825);
-INSERT INTO BENCH_COST VALUES('tcff','t100',4950,1,166650);
-INSERT INTO BENCH_COST VALUES('tcff','t500',124750,1,20833250);
-INSERT INTO BENCH_COST VALUES('tcff','t1k',499500,1,166666500);
-INSERT INTO BENCH_COST VALUES('tcff','t1k5',1124250,1,562499750);
-INSERT INTO BENCH_COST VALUES('tcff','t2k',1999000,1,1333333000);
-INSERT INTO BENCH_COST VALUES('tcff','v10',8194,9,8194);
-INSERT INTO BENCH_COST VALUES('tcff','v11',18434,10,18434);
-INSERT INTO BENCH_COST VALUES('tcff','v12',40962,11,40962);
-INSERT INTO BENCH_COST VALUES('tcff','v17',1966082,16,1966082);
-INSERT INTO BENCH_COST VALUES('tcff','v18',4194306,17,4194306);
-INSERT INTO BENCH_COST VALUES('tcff','w1k',1000000,1,1000000);
-INSERT INTO BENCH_COST VALUES('tcff','x10k',100020000,2,100020000);
-INSERT INTO BENCH_COST VALUES('tcff','y1k_4k',11998000,4000,11998000);
-INSERT INTO BENCH_COST VALUES('tcff','y1k_8k',39996000,8000,39996000);
-INSERT INTO BENCH_COST VALUES('tcff','y500_4k',9998000,4000,9998000);
-INSERT INTO BENCH_COST VALUES('tcff','y500_8k',35996000,8000,35996000);
--
GitLab